Comparison of Several Clustering Methods in Grouping Kale Landraces

Four clustering methods were compared for classification of a collection of 148 kale landraces ( Brassica oleracea L. acephala group) from northwestern Spain based on morphologic characters: the unweighted pair group method using arithmetic averages (UPGMA) and the Ward method, hierarchical cluster algorithms, and the modified location model (MLM) applied to both the UPGMA and the Ward method (UPGMA-MLM and Ward-MLM, respectively). Comparisons were based on five criteria and on subjective considerations about the structure of each method and the characteristics of the material evaluated. Although the UPGMA-MLM was superior according to the objective criteria, its slight advantage with respect to the Ward-MLM strategy did not overcome the fact that the initial UPGMA cluster generated a classification with little value. The Ward-MLM strategy generated five homogeneous groups with defined morphologic characteristics. Moreover, the Ward-MLM strategy allowed the identification of redundant landraces, which would permit the number of accessions in further critical trials to be reduced.